Tables and figures for publications
Alberto Garcia-Basteiro and Joe Brew
March 28, 2017
Tables
Descriptive tables
Global differences by:, age, sex, world region
| IHME | WHO | Dif | % difference (WHO / IHME) | % difference (IHME / WHO) | ||
|---|---|---|---|---|---|---|
| Total | HIV+TB only | 211604 | 389042 | 177438.13321 | 83.85% | -45.61% |
| TB only | 1111312 | 1379440 | 268128.44955 | 24.13% | -19.44% | |
| Total TB | 1322916 | 1768482 | 445566.58275 | 33.68% | -25.19% | |
| Adults | HIV+TB only | 177567 | 348026 | 170458.90473 | 96% | -48.98% |
| TB only | 1075691 | 1210620 | 134929.12946 | 12.54% | -11.15% | |
| Total TB | 1253257 | 1558645 | 305388.03419 | 24.37% | -19.59% | |
| Children | HIV+TB only | 34037 | 41016 | 6979.22848 | 20.5% | -17.02% |
| TB only | 35621 | 168821 | 133199.32009 | 373.93% | -78.9% | |
| Total TB | 69659 | 209837 | 140178.54857 | 201.24% | -66.8% | |
| Female | HIV+TB only | 78110 | 143496 | 65386.51804 | 83.71% | -45.57% |
| TB only | 367764 | 352488 | 15276.43876 | -4.15% | 4.33% | |
| Total TB | 445874 | 495984 | 50110.07929 | 11.24% | -10.1% | |
| Male | HIV+TB only | 99457 | 204471 | 105013.90757 | 105.59% | -51.36% |
| TB only | 707927 | 858132 | 150205.56821 | 21.22% | -17.5% | |
| Total TB | 807383 | 1062603 | 255219.47578 | 31.61% | -24.02% | |
| AMR | HIV+TB only | 579 | 620 | 41.31917 | 7.14% | -6.66% |
| TB only | 2036 | 1914 | 122.37010 | -6.01% | 6.39% | |
| Total TB | 2615 | 2534 | 81.05093 | -3.1% | 3.2% | |
| EMR | HIV+TB only | 165 | 533 | 368.30203 | 223.05% | -69.04% |
| TB only | 14658 | 14572 | 85.51575 | -0.58% | 0.59% | |
| Total TB | 14823 | 15106 | 282.78629 | 1.91% | -1.87% | |
| EUR | HIV+TB only | 212 | 374 | 161.68749 | 76.15% | -43.23% |
| TB only | 2383 | 2999 | 615.93832 | 25.85% | -20.54% | |
| Total TB | 2595 | 3373 | 777.62581 | 29.97% | -23.06% | |
| SEA | HIV+TB only | 19310 | 28870 | 9560.04060 | 49.51% | -33.11% |
| TB only | 333250 | 345889 | 12639.43214 | 3.79% | -3.65% | |
| Total TB | 352560 | 374759 | 22199.47275 | 6.3% | -5.92% | |
| WPR | HIV+TB only | 2057 | 2010 | 47.13348 | -2.29% | 2.35% |
| TB only | 39055 | 28351 | 10704.20283 | -27.41% | 37.76% | |
| Total TB | 41112 | 30361 | 10751.33632 | -26.15% | 35.41% |
Analytical tables
Table with model output for estimating likelihood or magnitude of difference in estimates by HIV, age, sex, and region.
This section is unfinished.
Graphs
Descriptive graphs
Standardized difference
Rankings of highest absolute and standardized differences for IHME and WHO.
Standardized difference for incidence
AB metric (a-b) / (a+b)
Rankings of highest absolute and standardized differences for IHME and WHO.
Maps
The below scatterplot shows the correlation between WHO (x-axis) estimates and IHME (y-axis) estimates, with each point colored by its (WHO-defined) region.
Analytical for adjusted_stand_dif
In the following four charts, Libya has been excluded as an outlier.
a) HIV prevalence among TB cases
b) Rifampicine resistance (MDR prevalence)
b.i) Rifampicine resistance (MDR prevalence using reported cases)
c) CDR (case detection rate per WHO)
c.i) CDR (case detection rate per IHME)
c.ii) Correlation between different case detections rates
d) CFR (case fatality rate)
Prevalence survey’s effect
Association of prevalence survey and standardized difference
Analytical for stand_dif_inc_adj
a) HIV prevalence among TB cases
b) Rifampicine resistance (MDR prevalence)
b.i) Rifampicine resistance (MDR prevalence using reported cases)
c) CDR (case detection rate per WHO)
Error: <text>:16:0: unexpected end of input
14: # Those countries with low case detection (oftentimes have prev survey),
15: # WHO estimates more deaths than IHME
^c.i) CDR (case detection rate per IHME)
d) CFR (case fatality rate)
Prevalence survey’s effect
Association of prevalence survey and standardized difference
Analytical for adjusted_ab
a) HIV prevalence among TB cases
b) Rifampicine resistance (MDR prevalence)
b.i) Rifampicine resistance (MDR prevalence using reported cases)
c) CDR (case detection rate per WHO)
c.i) CDR (case detection rate per IHME)
d) CFR (case fatality rate)
Prevalence survey’s effect
Association of prevalence survey and standardized difference
Modeling
Linear regression to estimate effect of prevalence survey on absolute difference in cases (WHO minus IHME), adjusting for region.
95% confidence intervals
Linear regression to estimate effect of prevalence survey on adjusted standardized difference in cases, adjusting for region.
95% confidence intervals
Interactive map
Relative difference (function of maximum estimate)
(Unfinished)
Map of world
Map of Mozambique
Correlation coefficients
Correlation of adjusted stand diff with a) HIV prevalence, CDR by both, CFR, MDR prevalence.
cor(df$adjusted_stand_dif, df$newrel_hivpos, use = 'complete.obs')[1] 0.09204849cor(df$adjusted_stand_dif, df$gb_c_cdr, use = 'complete.obs')[1] -0.3688292cor(df$adjusted_stand_dif, df$cdr_ihme, use = 'complete.obs')[1] 0.4283954cor(df$adjusted_stand_dif, df$case_fatality_rate_2014, use = 'complete.obs')[1] -0.1353054cor(df$adjusted_stand_dif, df$case_fatality_rate_2012_to_2014, use = 'complete.obs')[1] -0.1171681cor(df$adjusted_stand_dif, df$case_fatality_rate_2015, use = 'complete.obs')[1] -0.1586533cor(df$adjusted_stand_dif, df$case_fatality_rate_2014_new, use = 'complete.obs')[1] -0.1587409cor(df$adjusted_stand_dif, df$case_fatality_rate_2012_to_2014_new, use = 'complete.obs')[1] -0.1147299cor(df$adjusted_stand_dif, df$case_fatality_rate_2015_new, use = 'complete.obs')[1] -0.1760415cor(df$adjusted_stand_dif, df$case_fatality_rate_2015_adjusted, use = 'complete.obs')[1] -0.1825428cor(df$adjusted_stand_dif, df$p_mdr_new, use = 'complete.obs')[1] 0.05559062cor(df$adjusted_stand_dif, df$reported_mdr, use = 'complete.obs')[1] -0.0131539Hypothesis testing on prevalence survey
Does region affect likelihood of having a prevalence survey?
xt <- table(df$prevsurvey, df$who_region)
xt
AFR AMR EMR EUR SEA WPR
0 37 37 20 52 8 22
1 10 0 2 0 3 4chisq.test(xt)
Pearson's Chi-squared test
data: xt
X-squared = 21.511, df = 5, p-value = 0.0006482Does having a prev survey affect the adjusted stand diff?
t.test(x = df$adjusted_stand_dif[df$prevsurvey == 0],
y = df$adjusted_stand_dif[df$prevsurvey == 1])
Welch Two Sample t-test
data: df$adjusted_stand_dif[df$prevsurvey == 0] and df$adjusted_stand_dif[df$prevsurvey == 1]
t = -2.1643, df = 21.066, p-value = 0.04207
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
-39.2150725 -0.7865973
sample estimates:
mean of x mean of y
-4.545393 15.455442 Supplemental charts
April 18, 2018
Concordance between adult and child measures
Effect of prevalence survey
Is CDR associated with stand diff in the linear regression among those countries without prevalence survey?
x <- df
x$gb_c_cdr[!is.finite(x$gb_c_cdr)] <- NA
x$original_stand_dif[!is.finite(x$original_stand_dif)] <- NA
fit <- lm(gb_c_cdr ~ original_stand_dif, data = x[x$prevsurvey == 0,])
summary(fit)
Call:
lm(formula = gb_c_cdr ~ original_stand_dif, data = x[x$prevsurvey ==
0, ])
Residuals:
Min 1Q Median 3Q Max
-42.50 -12.33 4.16 10.80 23.21
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 76.0320 1.1661 65.201 < 0.0000000000000002 ***
original_stand_dif -0.3555 0.1123 -3.166 0.00187 **
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 14.35 on 151 degrees of freedom
(23 observations deleted due to missingness)
Multiple R-squared: 0.06226, Adjusted R-squared: 0.05605
F-statistic: 10.03 on 1 and 151 DF, p-value: 0.001868Yes. Among countries without a prevalence survey, a 1 unit increase in the standardized difference is associated with a 0.355 unit decrease in the CDR. This is significant (p < 0.01).